Hugging Face's logo

facebook
/
galactica-30b

Text Generation
Transformers
PyTorch
opt
galactica
text-generation-inference
Model card Files Community
12
mrm8488 commited on
Commit
31afe9c
•
1 Parent(s): 25362f0

Upload the model

Browse files
Files changed (9) hide show
  1. config.json +30 -0
  2. pytorch_model-00001-of-00007.bin +3 -0
  3. pytorch_model-00002-of-00007.bin +3 -0
  4. pytorch_model-00003-of-00007.bin +3 -0
  5. pytorch_model-00004-of-00007.bin +3 -0
  6. pytorch_model-00005-of-00007.bin +3 -0
  7. pytorch_model-00006-of-00007.bin +3 -0
  8. pytorch_model-00007-of-00007.bin +3 -0
  9. pytorch_model.bin.index.json +297 -0
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "activation_function": "gelu",
4
+ "architectures": [
5
+ "OPTModel"
6
+ ],
7
+ "attention_dropout": 0.1,
8
+ "bias": false,
9
+ "bos_token_id": 0,
10
+ "do_layer_norm_before": true,
11
+ "dropout": 0.1,
12
+ "eos_token_id": 2,
13
+ "ffn_dim": 28672,
14
+ "hidden_size": 7168,
15
+ "init_std": 0.02,
16
+ "layer_norm_elementwise_affine": false,
17
+ "layerdrop": 0.0,
18
+ "learned_embeddings": true,
19
+ "max_position_embeddings": 2048,
20
+ "model_type": "opt",
21
+ "num_attention_heads": 56,
22
+ "num_hidden_layers": 48,
23
+ "pad_token_id": 1,
24
+ "scale_embeddings": false,
25
+ "torch_dtype": "float16",
26
+ "transformers_version": "4.21.0.dev0",
27
+ "use_cache": true,
28
+ "vocab_size": 50000,
29
+ "word_embed_proj_dim": 7168
30
+ }
pytorch_model-00001-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1cad10954e544c44aabd29f31e67292d1bc819d2e7b9842f14fdcef88d58f93
3
+ size 9789123397
pytorch_model-00002-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b919005245e2b77d57bf3a73ac18415083aa32b6e2e4e89c96b8d988453a0e7f
3
+ size 9865018181
pytorch_model-00003-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30a274571d49a30bb4d6872e69b96ad191fa22c92427d160c74ce225a566bd71
3
+ size 9865018181
pytorch_model-00004-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05db345d4fcca580bed2c6e9d0fe8feead207c2c2fa8384c27c94cbd4ed0e0bf
3
+ size 9865018181
pytorch_model-00005-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa79446f17da0f3b9f8815a3628c2b1935936ec819f09a5865ce4e3c4ee51aa7
3
+ size 9865018181
pytorch_model-00006-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d6ce164b560f4601d48f61c2a8d598106faa9f4b89c39334a712429649b75c8
3
+ size 9865018181
pytorch_model-00007-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0379c39b5a0cb59453b14738ef1d4924e93599aba4e57f2599036e76f36532f6
3
+ size 822084583
pytorch_model.bin.index.json ADDED
@@ -0,0 +1,297 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 59936206848
4
+ },
5
+ "weight_map": {
6
+ "decoder.embed_positions.weight": "pytorch_model-00001-of-00007.bin",
7
+ "decoder.embed_tokens.weight": "pytorch_model-00001-of-00007.bin",
8
+ "decoder.layers.0.fc1.weight": "pytorch_model-00001-of-00007.bin",
9
+ "decoder.layers.0.fc2.weight": "pytorch_model-00001-of-00007.bin",
10
+ "decoder.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
11
+ "decoder.layers.0.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
12
+ "decoder.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
13
+ "decoder.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
14
+ "decoder.layers.1.fc1.weight": "pytorch_model-00001-of-00007.bin",
15
+ "decoder.layers.1.fc2.weight": "pytorch_model-00001-of-00007.bin",
16
+ "decoder.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
17
+ "decoder.layers.1.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
18
+ "decoder.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
19
+ "decoder.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
20
+ "decoder.layers.10.fc1.weight": "pytorch_model-00002-of-00007.bin",
21
+ "decoder.layers.10.fc2.weight": "pytorch_model-00002-of-00007.bin",
22
+ "decoder.layers.10.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
23
+ "decoder.layers.10.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
24
+ "decoder.layers.10.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
25
+ "decoder.layers.10.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
26
+ "decoder.layers.11.fc1.weight": "pytorch_model-00002-of-00007.bin",
27
+ "decoder.layers.11.fc2.weight": "pytorch_model-00002-of-00007.bin",
28
+ "decoder.layers.11.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
29
+ "decoder.layers.11.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
30
+ "decoder.layers.11.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
31
+ "decoder.layers.11.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
32
+ "decoder.layers.12.fc1.weight": "pytorch_model-00002-of-00007.bin",
33
+ "decoder.layers.12.fc2.weight": "pytorch_model-00002-of-00007.bin",
34
+ "decoder.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
35
+ "decoder.layers.12.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
36
+ "decoder.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
37
+ "decoder.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
38
+ "decoder.layers.13.fc1.weight": "pytorch_model-00002-of-00007.bin",
39
+ "decoder.layers.13.fc2.weight": "pytorch_model-00002-of-00007.bin",
40
+ "decoder.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
41
+ "decoder.layers.13.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
42
+ "decoder.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
43
+ "decoder.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
44
+ "decoder.layers.14.fc1.weight": "pytorch_model-00002-of-00007.bin",
45
+ "decoder.layers.14.fc2.weight": "pytorch_model-00002-of-00007.bin",
46
+ "decoder.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
47
+ "decoder.layers.14.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
48
+ "decoder.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
49
+ "decoder.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
50
+ "decoder.layers.15.fc1.weight": "pytorch_model-00003-of-00007.bin",
51
+ "decoder.layers.15.fc2.weight": "pytorch_model-00003-of-00007.bin",
52
+ "decoder.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
53
+ "decoder.layers.15.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
54
+ "decoder.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
55
+ "decoder.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
56
+ "decoder.layers.16.fc1.weight": "pytorch_model-00003-of-00007.bin",
57
+ "decoder.layers.16.fc2.weight": "pytorch_model-00003-of-00007.bin",
58
+ "decoder.layers.16.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
59
+ "decoder.layers.16.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
60
+ "decoder.layers.16.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
61
+ "decoder.layers.16.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
62
+ "decoder.layers.17.fc1.weight": "pytorch_model-00003-of-00007.bin",
63
+ "decoder.layers.17.fc2.weight": "pytorch_model-00003-of-00007.bin",
64
+ "decoder.layers.17.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
65
+ "decoder.layers.17.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
66
+ "decoder.layers.17.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
67
+ "decoder.layers.17.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
68
+ "decoder.layers.18.fc1.weight": "pytorch_model-00003-of-00007.bin",
69
+ "decoder.layers.18.fc2.weight": "pytorch_model-00003-of-00007.bin",
70
+ "decoder.layers.18.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
71
+ "decoder.layers.18.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
72
+ "decoder.layers.18.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
73
+ "decoder.layers.18.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
74
+ "decoder.layers.19.fc1.weight": "pytorch_model-00003-of-00007.bin",
75
+ "decoder.layers.19.fc2.weight": "pytorch_model-00003-of-00007.bin",
76
+ "decoder.layers.19.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
77
+ "decoder.layers.19.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
78
+ "decoder.layers.19.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
79
+ "decoder.layers.19.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
80
+ "decoder.layers.2.fc1.weight": "pytorch_model-00001-of-00007.bin",
81
+ "decoder.layers.2.fc2.weight": "pytorch_model-00001-of-00007.bin",
82
+ "decoder.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
83
+ "decoder.layers.2.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
84
+ "decoder.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
85
+ "decoder.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
86
+ "decoder.layers.20.fc1.weight": "pytorch_model-00003-of-00007.bin",
87
+ "decoder.layers.20.fc2.weight": "pytorch_model-00003-of-00007.bin",
88
+ "decoder.layers.20.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
89
+ "decoder.layers.20.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
90
+ "decoder.layers.20.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
91
+ "decoder.layers.20.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
92
+ "decoder.layers.21.fc1.weight": "pytorch_model-00003-of-00007.bin",
93
+ "decoder.layers.21.fc2.weight": "pytorch_model-00003-of-00007.bin",
94
+ "decoder.layers.21.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
95
+ "decoder.layers.21.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
96
+ "decoder.layers.21.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
97
+ "decoder.layers.21.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
98
+ "decoder.layers.22.fc1.weight": "pytorch_model-00003-of-00007.bin",
99
+ "decoder.layers.22.fc2.weight": "pytorch_model-00003-of-00007.bin",
100
+ "decoder.layers.22.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
101
+ "decoder.layers.22.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
102
+ "decoder.layers.22.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
103
+ "decoder.layers.22.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
104
+ "decoder.layers.23.fc1.weight": "pytorch_model-00004-of-00007.bin",
105
+ "decoder.layers.23.fc2.weight": "pytorch_model-00004-of-00007.bin",
106
+ "decoder.layers.23.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
107
+ "decoder.layers.23.self_attn.out_proj.weight": "pytorch_model-00003-of-00007.bin",
108
+ "decoder.layers.23.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
109
+ "decoder.layers.23.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
110
+ "decoder.layers.24.fc1.weight": "pytorch_model-00004-of-00007.bin",
111
+ "decoder.layers.24.fc2.weight": "pytorch_model-00004-of-00007.bin",
112
+ "decoder.layers.24.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
113
+ "decoder.layers.24.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
114
+ "decoder.layers.24.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
115
+ "decoder.layers.24.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
116
+ "decoder.layers.25.fc1.weight": "pytorch_model-00004-of-00007.bin",
117
+ "decoder.layers.25.fc2.weight": "pytorch_model-00004-of-00007.bin",
118
+ "decoder.layers.25.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
119
+ "decoder.layers.25.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
120
+ "decoder.layers.25.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
121
+ "decoder.layers.25.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
122
+ "decoder.layers.26.fc1.weight": "pytorch_model-00004-of-00007.bin",
123
+ "decoder.layers.26.fc2.weight": "pytorch_model-00004-of-00007.bin",
124
+ "decoder.layers.26.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
125
+ "decoder.layers.26.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
126
+ "decoder.layers.26.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
127
+ "decoder.layers.26.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
128
+ "decoder.layers.27.fc1.weight": "pytorch_model-00004-of-00007.bin",
129
+ "decoder.layers.27.fc2.weight": "pytorch_model-00004-of-00007.bin",
130
+ "decoder.layers.27.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
131
+ "decoder.layers.27.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
132
+ "decoder.layers.27.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
133
+ "decoder.layers.27.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
134
+ "decoder.layers.28.fc1.weight": "pytorch_model-00004-of-00007.bin",
135
+ "decoder.layers.28.fc2.weight": "pytorch_model-00004-of-00007.bin",
136
+ "decoder.layers.28.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
137
+ "decoder.layers.28.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
138
+ "decoder.layers.28.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
139
+ "decoder.layers.28.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
140
+ "decoder.layers.29.fc1.weight": "pytorch_model-00004-of-00007.bin",
141
+ "decoder.layers.29.fc2.weight": "pytorch_model-00004-of-00007.bin",
142
+ "decoder.layers.29.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
143
+ "decoder.layers.29.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
144
+ "decoder.layers.29.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
145
+ "decoder.layers.29.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
146
+ "decoder.layers.3.fc1.weight": "pytorch_model-00001-of-00007.bin",
147
+ "decoder.layers.3.fc2.weight": "pytorch_model-00001-of-00007.bin",
148
+ "decoder.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
149
+ "decoder.layers.3.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
150
+ "decoder.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
151
+ "decoder.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
152
+ "decoder.layers.30.fc1.weight": "pytorch_model-00004-of-00007.bin",
153
+ "decoder.layers.30.fc2.weight": "pytorch_model-00004-of-00007.bin",
154
+ "decoder.layers.30.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
155
+ "decoder.layers.30.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
156
+ "decoder.layers.30.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
157
+ "decoder.layers.30.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
158
+ "decoder.layers.31.fc1.weight": "pytorch_model-00005-of-00007.bin",
159
+ "decoder.layers.31.fc2.weight": "pytorch_model-00005-of-00007.bin",
160
+ "decoder.layers.31.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
161
+ "decoder.layers.31.self_attn.out_proj.weight": "pytorch_model-00004-of-00007.bin",
162
+ "decoder.layers.31.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
163
+ "decoder.layers.31.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
164
+ "decoder.layers.32.fc1.weight": "pytorch_model-00005-of-00007.bin",
165
+ "decoder.layers.32.fc2.weight": "pytorch_model-00005-of-00007.bin",
166
+ "decoder.layers.32.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
167
+ "decoder.layers.32.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
168
+ "decoder.layers.32.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
169
+ "decoder.layers.32.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
170
+ "decoder.layers.33.fc1.weight": "pytorch_model-00005-of-00007.bin",
171
+ "decoder.layers.33.fc2.weight": "pytorch_model-00005-of-00007.bin",
172
+ "decoder.layers.33.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
173
+ "decoder.layers.33.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
174
+ "decoder.layers.33.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
175
+ "decoder.layers.33.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
176
+ "decoder.layers.34.fc1.weight": "pytorch_model-00005-of-00007.bin",
177
+ "decoder.layers.34.fc2.weight": "pytorch_model-00005-of-00007.bin",
178
+ "decoder.layers.34.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
179
+ "decoder.layers.34.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
180
+ "decoder.layers.34.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
181
+ "decoder.layers.34.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
182
+ "decoder.layers.35.fc1.weight": "pytorch_model-00005-of-00007.bin",
183
+ "decoder.layers.35.fc2.weight": "pytorch_model-00005-of-00007.bin",
184
+ "decoder.layers.35.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
185
+ "decoder.layers.35.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
186
+ "decoder.layers.35.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
187
+ "decoder.layers.35.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
188
+ "decoder.layers.36.fc1.weight": "pytorch_model-00005-of-00007.bin",
189
+ "decoder.layers.36.fc2.weight": "pytorch_model-00005-of-00007.bin",
190
+ "decoder.layers.36.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
191
+ "decoder.layers.36.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
192
+ "decoder.layers.36.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
193
+ "decoder.layers.36.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
194
+ "decoder.layers.37.fc1.weight": "pytorch_model-00005-of-00007.bin",
195
+ "decoder.layers.37.fc2.weight": "pytorch_model-00005-of-00007.bin",
196
+ "decoder.layers.37.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
197
+ "decoder.layers.37.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
198
+ "decoder.layers.37.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
199
+ "decoder.layers.37.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
200
+ "decoder.layers.38.fc1.weight": "pytorch_model-00005-of-00007.bin",
201
+ "decoder.layers.38.fc2.weight": "pytorch_model-00005-of-00007.bin",
202
+ "decoder.layers.38.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
203
+ "decoder.layers.38.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
204
+ "decoder.layers.38.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
205
+ "decoder.layers.38.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
206
+ "decoder.layers.39.fc1.weight": "pytorch_model-00006-of-00007.bin",
207
+ "decoder.layers.39.fc2.weight": "pytorch_model-00006-of-00007.bin",
208
+ "decoder.layers.39.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
209
+ "decoder.layers.39.self_attn.out_proj.weight": "pytorch_model-00005-of-00007.bin",
210
+ "decoder.layers.39.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
211
+ "decoder.layers.39.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
212
+ "decoder.layers.4.fc1.weight": "pytorch_model-00001-of-00007.bin",
213
+ "decoder.layers.4.fc2.weight": "pytorch_model-00001-of-00007.bin",
214
+ "decoder.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
215
+ "decoder.layers.4.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
216
+ "decoder.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
217
+ "decoder.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
218
+ "decoder.layers.40.fc1.weight": "pytorch_model-00006-of-00007.bin",
219
+ "decoder.layers.40.fc2.weight": "pytorch_model-00006-of-00007.bin",
220
+ "decoder.layers.40.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
221
+ "decoder.layers.40.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
222
+ "decoder.layers.40.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
223
+ "decoder.layers.40.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
224
+ "decoder.layers.41.fc1.weight": "pytorch_model-00006-of-00007.bin",
225
+ "decoder.layers.41.fc2.weight": "pytorch_model-00006-of-00007.bin",
226
+ "decoder.layers.41.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
227
+ "decoder.layers.41.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
228
+ "decoder.layers.41.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
229
+ "decoder.layers.41.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
230
+ "decoder.layers.42.fc1.weight": "pytorch_model-00006-of-00007.bin",
231
+ "decoder.layers.42.fc2.weight": "pytorch_model-00006-of-00007.bin",
232
+ "decoder.layers.42.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
233
+ "decoder.layers.42.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
234
+ "decoder.layers.42.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
235
+ "decoder.layers.42.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
236
+ "decoder.layers.43.fc1.weight": "pytorch_model-00006-of-00007.bin",
237
+ "decoder.layers.43.fc2.weight": "pytorch_model-00006-of-00007.bin",
238
+ "decoder.layers.43.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
239
+ "decoder.layers.43.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
240
+ "decoder.layers.43.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
241
+ "decoder.layers.43.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
242
+ "decoder.layers.44.fc1.weight": "pytorch_model-00006-of-00007.bin",
243
+ "decoder.layers.44.fc2.weight": "pytorch_model-00006-of-00007.bin",
244
+ "decoder.layers.44.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
245
+ "decoder.layers.44.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
246
+ "decoder.layers.44.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
247
+ "decoder.layers.44.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
248
+ "decoder.layers.45.fc1.weight": "pytorch_model-00006-of-00007.bin",
249
+ "decoder.layers.45.fc2.weight": "pytorch_model-00006-of-00007.bin",
250
+ "decoder.layers.45.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
251
+ "decoder.layers.45.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
252
+ "decoder.layers.45.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
253
+ "decoder.layers.45.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
254
+ "decoder.layers.46.fc1.weight": "pytorch_model-00006-of-00007.bin",
255
+ "decoder.layers.46.fc2.weight": "pytorch_model-00006-of-00007.bin",
256
+ "decoder.layers.46.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
257
+ "decoder.layers.46.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
258
+ "decoder.layers.46.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
259
+ "decoder.layers.46.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
260
+ "decoder.layers.47.fc1.weight": "pytorch_model-00007-of-00007.bin",
261
+ "decoder.layers.47.fc2.weight": "pytorch_model-00007-of-00007.bin",
262
+ "decoder.layers.47.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
263
+ "decoder.layers.47.self_attn.out_proj.weight": "pytorch_model-00006-of-00007.bin",
264
+ "decoder.layers.47.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
265
+ "decoder.layers.47.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
266
+ "decoder.layers.5.fc1.weight": "pytorch_model-00001-of-00007.bin",
267
+ "decoder.layers.5.fc2.weight": "pytorch_model-00001-of-00007.bin",
268
+ "decoder.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
269
+ "decoder.layers.5.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
270
+ "decoder.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
271
+ "decoder.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
272
+ "decoder.layers.6.fc1.weight": "pytorch_model-00001-of-00007.bin",
273
+ "decoder.layers.6.fc2.weight": "pytorch_model-00001-of-00007.bin",
274
+ "decoder.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
275
+ "decoder.layers.6.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
276
+ "decoder.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
277
+ "decoder.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
278
+ "decoder.layers.7.fc1.weight": "pytorch_model-00002-of-00007.bin",
279
+ "decoder.layers.7.fc2.weight": "pytorch_model-00002-of-00007.bin",
280
+ "decoder.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
281
+ "decoder.layers.7.self_attn.out_proj.weight": "pytorch_model-00001-of-00007.bin",
282
+ "decoder.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
283
+ "decoder.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
284
+ "decoder.layers.8.fc1.weight": "pytorch_model-00002-of-00007.bin",
285
+ "decoder.layers.8.fc2.weight": "pytorch_model-00002-of-00007.bin",
286
+ "decoder.layers.8.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
287
+ "decoder.layers.8.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
288
+ "decoder.layers.8.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
289
+ "decoder.layers.8.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
290
+ "decoder.layers.9.fc1.weight": "pytorch_model-00002-of-00007.bin",
291
+ "decoder.layers.9.fc2.weight": "pytorch_model-00002-of-00007.bin",
292
+ "decoder.layers.9.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
293
+ "decoder.layers.9.self_attn.out_proj.weight": "pytorch_model-00002-of-00007.bin",
294
+ "decoder.layers.9.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
295
+ "decoder.layers.9.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin"
296
+ }
297
+ }